Mohammad Kazemi - Department of Statistics, Faculty of Mathematical Sciences, Shahrood University of Technology, Shahrood, Iran

In this paper, we propose a two-stage approach for feature selection in varying coefficient models with ultra-high-dimensional predictors. Specifically, we first employ partial correlation coefficient for screening, and then penalized rank regression is applied for dimension-reduced varying coefficient models to further select important predictors and estimate the coefficient functions. Simulation studies are carried out to examine the performance of proposed approach. We also illustrate it by a real data example.

Big data, feature screening, partial correlation, rank regression